Author Topic: [ZigBee] M256RFR2 methodology of increasing TX power (Read 2760 times)

danergo · « **on:** March 20, 2018, 10:17:48 am »

Hello, forumers,

I've built now a dozens of custom HWs of the ATMEGA256RFR2 and they all works quite okay (Coordinator, Routers, EDs).

However I usually experience signal losses between EDs and (R or C).
Now of course on the C I can put a nice big external antenna, however on the ED I use a chip antenna (http://uk.farnell.com/johanson-technology/2450at18a100e/antenna-ceramic-2-45ghz/dp/1885496).
My balun is: http://uk.farnell.com/johanson-technology/2450bm15a0015e/balun-imp-matched-2-4-2-5ghz-50ohm/dp/2341895

Now I checked the datasheet of the M256RFR2 and I read that it produce a max TX power of 3.5dBm (14.5mA) which is really small compared to for example:
JN5168-001-M05: 9.5dBm (35mA)
JN5168-001-M06: 22dBm (175mA) !!

Sure, I don't want to waste EDs batteries for TXing @22dBm. But my question still stands.

Is there any better alternative of the balun or chip antenna to achieve better results?

Is it better to use a separate small uC and a standalone RF chip which can achieve higher performance?

(Now the EDs usually disconnects from the C which is 8-10m far. There are walls and cables between them, and of course line of sight is also out of question.)

Thank you very much.

ataradov · « **Reply #1 on:** March 20, 2018, 05:18:01 pm »

The typical path is to use external PA/LNA. There is some chip from Skyworks that a lot of Atmel amplified design use. I believe it is still the best solution out there as far as amplifiers go.

Going with external radio is a bit extreme form a re-design point of view, but if it is something you can do, then it may be a solution as well.

danergo · « **Reply #2 on:** March 22, 2018, 07:31:50 am »

Thanks, that's what I looked for.

I found this: http://www.skyworksinc.com/uploads/documents/SKY66114_11_203678F.pdf

What do you think, if I have a C and 3 EDs, will it help if I use this FEM in only the C?
Then the C can TX@22dBm, and it can get a really sensitive input.

But I still won't give extra load on EDs' batteries.

ataradov · « **Reply #3 on:** March 22, 2018, 07:36:05 am »

Quote from: danergo on March 22, 2018, 07:31:50 am

What do you think, if I have a C and 3 EDs, will it help if I use this FEM in only the C?
Then the C can TX@22dBm, and it can get a really sensitive input.

You will see the benefit, for sure. Whether it will be sufficient in your case - hard to tell.

danergo · « **Reply #4 on:** March 22, 2018, 08:51:04 pm »

Yepp, sure.

One more thing:
This FEM goes directly between balun and antenna?

Thanks!

ataradov · « **Reply #5 on:** March 22, 2018, 09:13:25 pm »

Generally yes, but really, RF design is up to you, it is hard to tell what is going to work. You may not even need to use a balun if you just use positive output in an unbalanced way. You will lose half the power, but it may not matter, since you will have to do it anyway to not saturate the amplifier input.

danergo · « **Reply #6 on:** March 31, 2018, 04:28:23 pm »

Thank you, now I understand almost everything.

Please confirm my assumtions:

1.) On the "SKY66114-11" I only need two operation modes: Receive LNA, and Transmit high-power mode.
2.) Therefore I need to directly connect the uC's PF3 (DIG4) to SKY's CTX (Pin2). (All the other pins are either constant or don't care.)
3.) Which antenna is better to use? (RFP or RFN?) Is it okay to directly connect it to the SKY's RFIN? Without any resistors or capacitors?
4.) What should I do with the other RFx Pin? (Maybe connect it ground with 50R?)
5.) On the ANT output of SKY I shall connect my antenna just the same as I would connect it to the balun output (with / without resistors, caps)?
6.) In BitCloud: what is needed to change to use SKY as FEM?

Thank you very much!

ataradov · « **Reply #7 on:** March 31, 2018, 05:17:03 pm »

First of all, I want to say that I'm not an RF expert. I just picked up some stuff while working with this hardware, but I never actually designed anything myself.

1. Correct, if you don't care about sleep modes.
2. Correct. But I would bring out other DIG pins and SKY pins on test points on a prototype board, just in case.
3. It should not matter which pin to use, but I'm not sure about proper decoupling, you will have to figure it out yourself. I would try to find some examples of such use of balanced outputs. I would also at least put a series capacitor, you can always short it later.
4. Probably just ground.
5. Connect as described in the SKY documentation. For such high power you would need filtering and matching networks, otherwise you will have a hard time passing any regulation tests.
6. In the HAL there should be a setting names ENABLE_RT_TX_INDICATOR or something like this. You need to enable it and recompile the HAL.

danergo · « **Reply #8 on:** April 07, 2018, 04:37:34 pm »

Thank you very much ataradov!

Looking at an Atmel Application Note (http://ww1.microchip.com/downloads/en/AppNotes/Atmel-42140-WIRELESS-AT03188-Performance%20test%20evaluation%20with%20external%20FEM_Application-Note.pdf):
They are connecting an RCB256RFR2 board (https://fccid.io/VW4A091786/Schematics/Schematics-1866750.pdf) to an SE2431L (http://www.skyworksinc.com/uploads/documents/SE2431L_202410K.pdf).

There are only one balun between the two of them. No resistors, no capacitors.
I'm also not an RF expert but somehow I don't feel this right.

Don't you remember where did you see these schematics? I've searched through everything but I never found anything that connects an differential RF to a Front-End-Module.
Also, in the balun's datasheet they states it is creating an 50 Ohm matched signal, and the FEM module requires the very same signal. (50 Ohm RF source).

I just stucked and have no clue where to head now.

ataradov · « **Reply #9 on:** April 07, 2018, 04:44:56 pm »

I would absolutely trust schematics you see from Atmel. Those things were done be very capable engineers. And SE2431L is the chip I originally thought about.

It is possible that the design I'm thinking about was never public. But there must be a general theory on this, not specifically for Atmel chips.

Anyway, give me some time, I'll try to find a person that designed all those things and ask him. That will only happen next week.

danergo · « **Reply #10 on:** April 07, 2018, 05:41:16 pm »

Thank you very much!

Your help is really appreciated!

danergo · « **Reply #11 on:** April 09, 2018, 06:06:06 am »

Hello,

I've also created a stackexchange topic for this issue.
Just wanted to link it here. It may reach some more people.

https://electronics.stackexchange.com/questions/367020/unbalanced-rf-output-to-fem-module-with-balanced-input

ataradov · « **Reply #12 on:** April 09, 2018, 05:02:55 pm »

Ok, I found the document that describes a design like this. Look for "Atmel AVR2080: REB231FE2 – Hardware User’s Manual".

Here is the schematic taken from it. The document goes into details on the design.

ataradov · « **Reply #13 on:** April 09, 2018, 05:13:41 pm »

One more thing to consider - calculate your power budget. Even with this configuration, transceiver output may be too hot, and internal attenuation may be too coarse. You may want to implement external attenuator, or at least provision places for one.

Otherwise you may run into regulatory issues trying to get under the limit, but not by a lot.

danergo · « **Reply #14 on:** April 09, 2018, 08:21:18 pm »

ataradov, that's fantastic!!!
Thank you very much!

-
Yes, I'll take care of the regulations just I did not want to manufacture PCBs which will be landed in the trash because of uncertain details.
Thanks again I think it will help me a lot.

danergo · « **Reply #15 on:** April 09, 2018, 09:35:23 pm »

In this datasheet I can see these lines (page 9):

"Avoiding a balun helps minimizing the bill of materials cost. In transmit mode, the
AT86RF231 transmit output power needs to be set higher compared to a differential
TRX-FEM interface using a balun."

How is it possible to "set higher" the output power?
It seems from this single sentence, that a differential TRX-FEM interface with a balun can also be a way, no?

Thanks!

ataradov · « **Reply #16 on:** April 09, 2018, 10:13:04 pm »

Well, the option with a balun is the obvious one. I thought the goal was to avoid having a balun. If you can use a balun, then what is the problem?

By "set higher" they mean if you can do so. By using single-ended mode you will lose 3 dB of output power right away. But that's not a problem, since you will likely have to scale TX power even further to meed PA input power requirements.

danergo · « **Reply #17 on:** April 10, 2018, 05:45:34 am »

Yes, now I got the balun vs no-balun situation.

Some days ago you wrote that using the balun might saturate the PA input in the FEM and I thought that I have to avoid the balun because of this reason.
Now of course with the balun and this FEM I could reach huge distances so might be a good idea to not use the balun because it saves cost and will make my circuit to fit more into regulation limits.

Quote from: ataradov on March 22, 2018, 09:13:25 pm

Generally yes, but really, RF design is up to you, it is hard to tell what is going to work. You may not even need to use a balun if you just use positive output in an unbalanced way. You will lose half the power, but it may not matter, since you will have to do it anyway to not saturate the amplifier input.

Anyway, thank you very much!

ataradov · « **Reply #18 on:** April 10, 2018, 05:47:39 am »

Yes, even without the balun, you will probably saturate the PA if you go at it with full output power. Or better yet, consider a PA with less amplification, which will also simplify power supply and will make it easy to meet regulations.

danergo · « **Reply #19 on:** April 10, 2018, 03:07:27 pm »

Yes, thanks!

Let me ask one more generic question here:
In AVR2080 both FEM and Atmel chips is powered from 3V.

However in the FEM's datasheet I see it is very well recommended to power the FEM with 1.8V.
That means no problem, but can I also connect directly those 2 components if the Atmel works at 3.3V and the FEM at 1.8V?

Or should I use the same power for the FEM as the Atmel and then change the Atmel's TX power to adjust a nice output which meets the regulations?

ataradov · « **Reply #20 on:** April 10, 2018, 03:19:10 pm »

Where do you see this recommendation? To get maximum power you must power it at 3.6V, and the lower the voltage, the lower the output power, which gets down to 17 dBm at 2.0V.

danergo · « **Reply #21 on:** April 10, 2018, 03:24:30 pm »

Sorry, I did mean no recommendation but sample.

On page 3 and 4 everywhere VCC1 is considered 1.8V.
(Table4, note 2-4; Table5)

Okay, in table 3 I can see that VCC1 can be between 1.7 and 3.6V, but there are no calculations or diagram which I could guess what happens if VCC1 = 3.3V.

ataradov · « **Reply #22 on:** April 10, 2018, 03:31:49 pm »

Ok, we are talking about different chips. I was looking at SE2431L, the one used on the reference design.

For SKY66114 VCC1 is a digital logic power, you can have any voltage compatible with the rest of your system. What happens what you vary VCC2 is listed in the Table 5 (Pout parameter). At 1.8V you only get +19 dB of amplification (vs +23 dB at 3.0V).

danergo · « **Reply #23 on:** April 10, 2018, 03:54:29 pm »

Oh, thanks.

That part was missing.
It is very misleading to me to include a parameter into an item if the parameter does not matter for the current specification item.

Last thing to here: from BitCloud's point of view, what is the recommended way to set the TX output power?
I guess setting PHY_TX_PWR manually would make things happen, however there might be better alternatives, such as CS_WriteParameter maybe?

ataradov · « **Reply #24 on:** April 10, 2018, 04:01:09 pm »

Quote from: danergo on April 10, 2018, 03:54:29 pm

Last thing to here: from BitCloud's point of view, what is the recommended way to set the TX output power?
I guess setting PHY_TX_PWR manually would make things happen, however there might be better alternatives, such as CS_WriteParameter maybe?

It does not matter. But changing a number in a config is the easiest way.


EEVblog Main Site	EEVblog on Youtube	EEVblog on Twitter	EEVblog on Facebook	EEVblog on Odysee

Author Topic: [ZigBee] M256RFR2 methodology of increasing TX power (Read 2760 times)

Share me